dalab / deep-ed

Source code for the EMNLP'17 paper "Deep Joint Entity Disambiguation with Local Neural Attention", https://arxiv.org/abs/1704.04920
Apache License 2.0
224 stars 50 forks source link

stack traceback: [C]: in function 'assert' entities/learn_e2v/minibatch_a.lua:53: in function 'process_one_line' #16

Open Jorigorn opened 5 years ago

Jorigorn commented 5 years ago

In this step 14 CUDA_VISIBLE_DEVICES=0 th entities/learn_e2v/learn_a.lua -root_data_dir $DATA_PATH |& tee log_train_entity_vecs

The last saved model is ent_vecs__ep_63.t7

% ll -h deep_ed_data/generated/ent_vecs/ total 6,5G drwxrwxr-x 2 zchen zchen 4,0K Nov 8 15:10 . drwxrwxr-x 5 zchen zchen 4,0K Nov 8 19:29 .. -rw-rw-r-- 1 zchen zchen 316M Nov 7 23:26 ent_vecsep_12.t7 -rw-rw-r-- 1 zchen zchen 316M Nov 8 00:38 ent_vecsep_15.t7 -rw-rw-r-- 1 zchen zchen 316M Nov 8 01:33 ent_vecsep_18.t7 -rw-rw-r-- 1 zchen zchen 316M Nov 8 02:28 ent_vecsep_21.t7 -rw-rw-r-- 1 zchen zchen 316M Nov 8 03:23 ent_vecsep_24.t7 -rw-rw-r-- 1 zchen zchen 316M Nov 8 04:18 ent_vecsep_27.t7 -rw-rw-r-- 1 zchen zchen 316M Nov 8 05:14 ent_vecsep_30.t7 -rw-rw-r-- 1 zchen zchen 316M Nov 8 06:09 ent_vecsep_33.t7 -rw-rw-r-- 1 zchen zchen 316M Nov 8 07:04 ent_vecsep_36.t7 -rw-rw-r-- 1 zchen zchen 316M Nov 8 07:59 ent_vecsep_39.t7 -rw-rw-r-- 1 zchen zchen 316M Nov 7 19:18 ent_vecsep_3.t7 -rw-rw-r-- 1 zchen zchen 316M Nov 8 08:54 ent_vecsep_42.t7 -rw-rw-r-- 1 zchen zchen 316M Nov 8 09:50 ent_vecsep_45.t7 -rw-rw-r-- 1 zchen zchen 316M Nov 8 10:56 ent_vecsep_48.t7 -rw-rw-r-- 1 zchen zchen 316M Nov 8 12:03 ent_vecsep_51.t7 -rw-rw-r-- 1 zchen zchen 316M Nov 8 13:09 ent_vecsep_54.t7 -rw-rw-r-- 1 zchen zchen 316M Nov 8 13:52 ent_vecsep_57.t7 -rw-rw-r-- 1 zchen zchen 316M Nov 8 14:30 ent_vecsep_60.t7 -rw-rw-r-- 1 zchen zchen 316M Nov 8 15:10 ent_vecsep_63.t7 -rw-rw-r-- 1 zchen zchen 316M Nov 7 20:46 ent_vecsep_6.t7 -rw-rw-r-- 1 zchen zchen 316M Nov 7 22:06 ent_vecs__ep_9.t7

finished with error: log:

/home/zchen/torch/install/bin/luajit: entities/learn_e2v/minibatch_a.lua:53: 2321219 Team Penske Can-Am a Blue Sunoco 1967 Camaro driven by Mark Donohue in this series designed for Pony cars like the Ford Mustang Penske entered Camaros won the series championship in 1968 and 1969 Later they switched to a red white blue American Motors backed 1970 AMC Javelin and later the restyl2323747 Paramarines Marine Raiders in May 1942 next to Camp Kearny moving to purpose built accommodation nearby at Camp Gillespie in September 1942 A second training camp and parachute training school opened at Hadnot Point on the New River in North Carolina in June 1942 but closed in July 1943 The 1st Parachute Battalion was attached to the 1st Marine Division for the invasion of Guadalcanal On 7 August 1942 the unit conducted an amphibious assault on the small island of Gavutu and later seized the neighbouring island of Tanambogo with other Marine units The battalion later moved to Guadalcanal fighting alongside the 1st in the Tasimboko raid and the Battle of Edson s Ridge The high casualties suffered by the Marine paratroopers led the battalion to be moved to Camp Kiser in Tontouta New Caledonia in September The 2nd Parachute Battalion performed a diversionary raid on Choiseul Island in October 1943 and later joined the 1st and 3rd Parachute Battalion on Bougainville The three parachute battalions with approximately 3 000 members had become the 1st Marine Parachute Regiment of the I Marine Amphibious Corps However the need for and cost of a parachute corps in the Marines was questioned as were other specialized CANDIDATES 367878,1.000,Marine Raiders 23620074,0.565,Marine Raiders (film) 32928286,0.002,Florida Marine Raiders GT: 1,367878,1.000,Marine Raiders stack traceback: [C]: in function 'assert' entities/learn_e2v/minibatch_a.lua:53: in function 'process_one_line' entities/learn_e2v/batch_dataset_a.lua:70: in function 'get_minibatch' entities/learn_e2v/learn_a.lua:187: in function 'train_ent_vecs' entities/learn_e2v/learn_a.lua:272: in main chunk [C]: in function 'dofile' ...chen/torch/install/lib/luarocks/rocks/trepl/scm-1/bin/th:150: in main chunk [C]: at 0x00405d50

Now still can training the model Step 15 with: CUDA_VISIBLE_DEVICES=0 th ed/ed.lua -root_data_dir $DATA_PATH -ent_vecs_filename $ENTITY_VECS -model 'global' |& tee log_train_ed

But have to comment the two lines assert code , as written in issue #12 Maybe this is because the error caused by the step before?

John Thanks a lot. :)